Changeset 14961

Timestamp:

2008-02-12T15:17:47+13:00 (16 years ago)

Author:

davidb

Message:

Setting filename metadata (Source) in BasPlug.pm looks to user its locale, in the first instance, to resolve what character encoding the file system uses.

File:

• gsdl/trunk/perllib/plugins/BasPlug.pm (modified) (2 diffs)

gsdl/trunk/perllib/plugins/BasPlug.pm

@@ -797 +797 @@
 
 
+sub filename_to_metadata
+{
+    my $self = shift (@_);  
+    my ($file, $encoding) = @_;
+
+    my $outhandle = $self->{'outhandle'};
+
+    my $filesystem_encoding = undef;
+
+    eval {
+    use POSIX qw(locale_h);
+
+    # With only one parameter, setlocale retrieves the current value
+    my $current_locale = setlocale(LC_CTYPE);
+
+    if ($current_locale =~ m/^.*\.(.*?)$/) {
+        my $char_encoding = lc($1);
+        $char_encoding =~ s/-/_/g;
+        $char_encoding =~ s/^utf_8$/utf8/;
+
+        if ($char_encoding =~ m/^\d+$/) {
+        if (defined $encodings::encoding->{"windows_$char_encoding"}) {
+            $char_encoding = "windows_$char_encoding";
+        }
+        elsif (defined $encodings::encoding->{"dos_$char_encoding"}) {
+            $char_encoding = "dos_$char_encoding";
+        }
+        }
+
+        if (($char_encoding =~ m/(?:ascii|utf8|unicode)/) 
+        || (defined $encodings::encoding->{$char_encoding})) {
+        $filesystem_encoding = $char_encoding;
+        }
+        else {
+        print $outhandle "Warning: Unsupported character encoding '$char_encoding' from locale '$current_locale'\n";
+        }
+    }
+
+
+    };
+    if ($@) {
+    print $outhandle "$@\n";
+    print $outhandle "Warning: Unable to establish locale.  Will assume filesytem is UTF-8\n";
+    
+    }
+    
+    my ($filemeta) = $file =~ /([^\\\/]+)$/;
+    
+    # how do we know what encoding the filename is in?
+    # => one answer is to check the locale
+
+    if (defined $filesystem_encoding) {
+    if ($filesystem_encoding !~ /(?:ascii|utf8|unicode)/) {
+        $filemeta = unicode::unicode2utf8(
+            unicode::convert2unicode($filesystem_encoding, \$filemeta)
+              );
+    }
+    }
+    # assume it is in the same encoding as its contents
+    elsif ((defined $encoding) && ($encoding !~ /(?:ascii|utf8|unicode)/)) {
+    $filemeta = unicode::unicode2utf8(
+        unicode::convert2unicode($encoding, \$filemeta)
+    );
+    }
+    
+    my $dmsafe_filemeta = &ghtml::dmsafe($filemeta);
+
+    return $dmsafe_filemeta;
+}
+
 
 # The BasPlug read_into_doc_obj() function. This function does all the
@@ -835 +905 @@
     # create a new document
     my $doc_obj = new doc ($filename, "indexed_doc");
+    my $top_section = $doc_obj->get_top_section();
+
     $doc_obj->set_OIDtype ($processor->{'OIDtype'}, $processor->{'OIDmetadata'});
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Language", $language);
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Encoding", $encoding);
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "FileSize", (-s $filename));
-
-    my ($filemeta) = $file =~ /([^\\\/]+)$/;
-    # how do we know what encoding the filename is in?
-    # assume it is in the same encoding as its contents
-    if ($encoding !~ /(?:ascii|utf8|unicode)/) {
-    $filemeta = unicode::unicode2utf8(
-        unicode::convert2unicode($encoding, \$filemeta)
-    );
-    }
-    $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Source", &ghtml::dmsafe($filemeta));
+    $doc_obj->add_utf8_metadata($top_section, "Language", $language);
+    $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
+    $doc_obj->add_utf8_metadata($top_section, "Plugin", "$self->{'plugin_type'}");
+    $doc_obj->add_utf8_metadata($top_section, "FileSize", (-s $filename));
+
+    my $filemeta = $self->filename_to_metadata($file,$encoding);
+    $doc_obj->add_utf8_metadata($top_section, "Source", $filemeta);
     if ($self->{'cover_image'}) {
     $self->associate_cover_image($doc_obj, $filename);